using sed how to append N number of lines

Question

I have the following csv file:

hostname1,this is a test,001^M
hostname2,this is
a test,002^M
hostname3,this
is
a
test,003^M

The first line is the proper format of the line. I want the other lines to follow the same format. The problem is for the second column, some of them have carriage returns so the third column is on a new line, but I want to use sed to join all the lines regardless of how many lines they broke apart.

Answer 1

Assuming the ^M sequences represent CR characters, i.e.

$ cat -et file.csv
hostname1,this is a test,001^M$
hostname2,this is$
a test,002^M$
hostname3,this$
is$
a$
test,003^M$

then you can join lines to the next if the do not end in CR using

sed -e :a -e '/\r$/!N; s/\n//; ta' file.csv

ex.

$ sed -e :a -e '/\r$/!N; s/\n//; ta' file.csv | cat -et
hostname1,this is a test,001^M$
hostname2,this isa test,002^M$
hostname3,thisisatest,003^M$